Machine Learning in Falls Prediction; A cognition-based predictor of falls for the acute neurological in-patient population

📝 Abstract

Background Information: Falls are associated with high direct and indirect costs, and significant morbidity and mortality for patients. Pathological falls are usually a result of a compromised motor system, and/or cognition. Very little research has been conducted on predicting falls based on this premise. Aims: To demonstrate that cognitive and motor tests can be used to create a robust predictive tool for falls. Methods: Three tests of attention and executive function (Stroop, Trail Making, and Semantic Fluency), a measure of physical function (Walk-12), a series of questions (concerning recent falls, surgery and physical function) and demographic information were collected from a cohort of 323 patients at a tertiary neurological center. The principal outcome was a fall during the in-patient stay (n = 54). Data-driven, predictive modelling was employed to identify the statistical modelling strategies which are most accurate in predicting falls, and which yield the most parsimonious models of clinical relevance. Results: The Trail test was identified as the best predictor of falls. Moreover, addition of any others variables, to the results of the Trail test did not improve the prediction (Wilcoxon signed-rank p < .001). The best statistical strategy for predicting falls was the random forest (Wilcoxon signed-rank p < .001), based solely on results of the Trail test. Tuning of the model results in the following optimized values: 68% (+- 7.7) sensitivity, 90% (+- 2.3) specificity, with a positive predictive value of 60%, when the relevant data is available. Conclusion: Predictive modelling has identified a simple yet powerful machine learning prediction strategy based on a single clinical test, the Trail test. Predictive evaluation shows this strategy to be robust, suggesting predictive modelling and machine learning as the standard for future predictive tools.

💡 Analysis

Background Information: Falls are associated with high direct and indirect costs, and significant morbidity and mortality for patients. Pathological falls are usually a result of a compromised motor system, and/or cognition. Very little research has been conducted on predicting falls based on this premise. Aims: To demonstrate that cognitive and motor tests can be used to create a robust predictive tool for falls. Methods: Three tests of attention and executive function (Stroop, Trail Making, and Semantic Fluency), a measure of physical function (Walk-12), a series of questions (concerning recent falls, surgery and physical function) and demographic information were collected from a cohort of 323 patients at a tertiary neurological center. The principal outcome was a fall during the in-patient stay (n = 54). Data-driven, predictive modelling was employed to identify the statistical modelling strategies which are most accurate in predicting falls, and which yield the most parsimonious models of clinical relevance. Results: The Trail test was identified as the best predictor of falls. Moreover, addition of any others variables, to the results of the Trail test did not improve the prediction (Wilcoxon signed-rank p < .001). The best statistical strategy for predicting falls was the random forest (Wilcoxon signed-rank p < .001), based solely on results of the Trail test. Tuning of the model results in the following optimized values: 68% (+- 7.7) sensitivity, 90% (+- 2.3) specificity, with a positive predictive value of 60%, when the relevant data is available. Conclusion: Predictive modelling has identified a simple yet powerful machine learning prediction strategy based on a single clinical test, the Trail test. Predictive evaluation shows this strategy to be robust, suggesting predictive modelling and machine learning as the standard for future predictive tools.

📄 Content

Machine Learning in Falls Prediction; A cognition-based predictor of falls for the acute neurological in-patient population

Mr. B.A. Mateen1

Mr. M. Bussas2

Dr. C. Doogan 3

Dr. D. Waller4

Dr. A. Saverino5

Dr. F. J. Király2,6*

Prof. E. D. Playford3,7

1 University College London, London, UK 2 Department of Statistical Science, University College London, London, UK
3 Therapy and Rehabilitation Services, National Hospital for Neurology & Neurosurgery, London, UK 4 Neurorehabilitation Unit, National Hospital for Neurology and Neurosurgery, London, UK 5 Wolfson Neuro Rehabilitation Centre, St Georges Hospital, London, UK 6 The Alan Turing Institute, London, UK 7 Institute of Neurology, University College London, London, UK

Author for correspondence (*):
Dr. Franz J. Király Department of Statistical Science,
University College London,
 Gower Street
London WC1E 6BT
United Kingdom

 Tel.: +44 - 20 - 7679 1259
Fax.: +44 - 20 - 3108 3105
 E-mail: f.kiraly@ucl.ac.uk

Declarations – Dr Diane Playford was supported by the National Institute for Health Research University College London Hospitals Biomedical Research Centre. The authors have no conflict of interests to declare. This study received no funding.

Contributions –

BAM and EDP conceived and planned the study with contribution from CD, DW and AS. BAM collected the data under the supervision of DW and EDP. Statistical analysis conducted by MB under the supervision of FJK, with contributions from BAM. Manuscript written by BAM and FJK, under the supervision of EDP.

2

Abstract Background Information – Falls are associated with high direct and indirect costs, and significant morbidity and mortality for patients. Pathological falls are usually a result of a compromised motor system, and/or cognition. Very little research has been conducted on predicting falls based on this premise.

Aims – To demonstrate that cognitive and motor tests can be used to create a robust predictive tool for falls.

Methods – Three tests of attention and executive function (Stroop, Trail Making, & Semantic Fluency), a measure of physical function (Walk-12), a series of questions (concerning recent falls, surgery and physical function) and demographic information were collected from a cohort of 323 patients at a tertiary neurological center. The principal outcome was a fall during the in-patient stay (n = 54). Data-driven, predictive modelling was employed to identify the statistical modelling strategies which are most accurate in predicting falls, and which yield the most parsimonious models of clinical relevance.

Results – The Trail test was identified as the best predictor of falls. Moreover, addition of any others variables, to the results of the Trail test did not improve the prediction (Wilcoxon signed-rank p < .001). The best statistical strategy for predicting falls was the random forest (Wilcoxon signed-rank p < .001), based solely on results of the Trail test. Tuning of the model results in the following optimized values: 68% (± 7.7) sensitivity, 90% (± 2.3) specificity, with a positive predictive value of 60%, when the relevant data is available.

Conclusion – Predictive modelling has identified a simple yet powerful machine learning prediction strategy based on a single clinical test, the Trail test. Predictive evaluation shows this strategy to be robust, suggesting predictive modelling and machine learning as the standard for future predictive tools.

Words - 284

3 Introduction The Cost and Prevalence of Falls, and Falls-related Injury Falls are a serious public health concern1, with potentially fatal consequences2, and significant financial implications for individuals, and their families3. In a single year in the USA, there were more than 10,000 fatal falls in the elderly population, and an additional 2.6 million medically treated falls-related injuries that were non-fatal, which resulted in a direct cost of close to US $20 billion4. In the UK, falls account for over 60% of all hospital in- patient related safety incidents5, resulting in an annual direct cost of £15 million6, on top of the billions already spent on treating falls-related injuries in the community that result in hospital admissions7,8. Some argue that if steps are not taken to address this problem, by the year 2030 the number of injuries resulting from falls will have increased by 100%9, therefore it is vital that steps are taken to prevent this astronomical rise in cost and harm to all the relevant stakeholders.

Clinical Relevance of Predicting Falls The current debate in the falls literature is whether probabilistic prediction is clinically useful, and whether it is more important than targeting modifiable risk factors10. We argue that these two approaches are not mutually exclusive, rather, making sound predictions is actually necessary for planning and evaluating interventions of any

This content is AI-processed based on ArXiv data.